Hi Paul
Thanks for the input :)
On 12 May 2011 09:56, Paul Merlin <eskatos_at_n0pe.org> wrote:
> Quoting "Manfred Riem" <mriem_at_manorrock.org>:
> > I am not saying it is easy to fix. However the set of libraries that I am
> > using is not causing any problems with regard to PermGen space, so your
> set of
> > libraries is causing the problem. Not the application server, and it is
> > very hard for the application server to solve the problem.
>
>
> IMHO the container *cannot* solve the problem because it's totally use-case
> dependant. The hypothetical solution would be different each time a library
> introduce the issue.
>
Then your suggestion of a health monitor needs to be implemented on a IDE
level or as a GlassFish watch dog layer ?
>
>
> > Hence my admonition to investigate which library is causing it.
>
> I would definetely go that way and either push a fix upstream or eradicate
> this
> library from my toolset.
>
In most projects I develop, (and I think other people develop) I have 10+
open source misc libraries and actually it relies on the Java eco system.
In practical terms checking each one would defeat the purpose of using the
Java eco system and actually Java. That's why I think it won't work for me.
>
> If you're stuck with this library you could do something hardcore like
> "kill"
> the incriminated references using reflection on application undeployment.
> But
> this means you know how to find them and you're doing something ilegal in
> JEE
> terms.
>
Yes, I don't think that's a good idea :)
regards
Richard.